Where and When Should IoT Data Deliver the Goods?
June 29, 2016 Leave a comment
• Where and how quickly do companies analyze IoT data? Do they iteratively push device data into a central warehouse en mass for analysis later, or do they process all of that data at or close to the source in real-time?
• It turns out that enterprises want answers not just here and now but also later and in greater detail, making the case for combined distributed and centralized data processing.
I maintain a friendly but superficial relationship with math, but I understand enough to admire ideas like Heisenberg’s uncertainty principle and Erwin Schrodinger’s related and now famous thought experiment about the wellbeing of secretly imprisoned felines. It’s intriguing to think that for certain pairs of physical properties, like both the location and velocity of a given particle, you can calculate a particle’s speed, but in so doing you forfeit the ability to also know its location.
Hmmm. That sounds a lot like a problem specific to the Internet of Things (IoT).
What if you want to understanding both where and when business analyze the seemingly boundless scope of data streaming off of instrumented devices? Are they in compliance with the uncertainty principle and either a) iteratively pushing device data into a central warehouse en masse for analysis later, or b) processing all of that data at or close to the source in real-time? The results may surprise you.
Apparently Heisenberg’s uncertainty principle does not apply to IoT data processing and analysis, at least in terms of what enterprise buyers want. Hint: they want both. Our recent survey of more than 1,000 IT buyers (entitled Enterprise IoT: Global Investment Survey & Insights) showed that IT professionals and business owners do not prefer velocity over location or location over velocity. Our study revealed that 42% of IoT practitioners (planned or current) analyze IoT continuously, while it is in motion (See Figure 1), nearly doubling those who opted to instead to analyze data at rest.
Interestingly and seemingly in contradiction to the velocity of IoT analysis (Figure 1), our study also showed that (47%) of IoT buyers preferred to centralize and analyze IoT data within a data warehouse (See Figure 2).
This indicates a preference for what can be considered an optimal IoT deployment model wherein companies push their IoT data in real-time to a centralized data warehouse where it is then processed and combined with systems of record (ERP solutions, sales enablement software, etc.) and finally analyzed within that broader context.
These findings uncover an important truth. The cat is both alive and dead (sorry kitty). Velocity and location are both inexorably intertwined and should not be considered separately from one another. Vendors aspiring to deliver a solution-complete IoT platform must take this into account. It is not enough to simply aggregate operational and business analytics back in the data center within the safe confines of a data warehouse. That is of course a necessary component as it allows enterprise customers to create a single, 360 degree view of their business by blending sensor data with line of business information. What’s more important is the capacity to allow for the analysis of data both in real-time and asynchronously, and both centrally and at the edge (close to or on the IoT device).